Internships

HealthCubed : Smart Diagnostics Starts Here

HealthCubed is a healthcare technology company focused on providing innovative solutions for underserved areas. They have built a portable diagnostic device designed to perform a wide range of medical tests, including measuring vital signs like blood pressure, blood glucose levels, cholesterol levels, and more. This device is designed to provide quick and reliable health assessments in areas with limited healthcare infrastructure. Over the course of two weeks, I had the opportunity to work as a part of a team to reduce cost for HealthCubed devices by building ECG circuits with more efficient microcontrollers and IC chips.

The internship commenced with an extensive demonstration of their multifunctional device. This device showcased the remarkable fusion of electronics, power electronics, electrical engineering, and high-voltage electrical principles. I dissected the rudimentary components, such as transistors, illustrating their role as electronic signal switches or gates. Logic gates, Boolean algebra, and the amplification of electronic signals, as exemplified by FM radio, further enriched the comprehension of these fundamental building blocks.

Then, I had hands-on immersion into electronics, employing tools like multimeters, breadboards, resistors, and wires. NPN and PNP transistors were examined, alongside the utilization of ceramic and electrolytic capacitors. A practical exercise involved constructing circuits with two resistors in series and measuring the input and output voltages. The relationship between these voltages was elegantly encapsulated in the equation Vout = Vin * [R2/(R1+R2)]. The exploration then expanded to include the usage of a digital oscilloscope, which enabled the visualization and measurement of voltage in an ECG circuit. The incorporation of components like ESP32 microcontroller, IC chips, master and slave boards, MISO and MOSI pins showcased the integration of theoretical knowledge into practical implementations.

Further, I dived deep into memory hierarchies encompassing RAM, ROM, primary, and secondary storage proved insightful. Concepts of digital and analog input/output, ADC, DAC, registers, and data sheets were elucidated with the aid of platforms like Arduino, ESP32, and STM32. The instrumental role of evaluation boards like Nucleo and Eval in facilitating prototyping and development was highlighted. This day also illuminated the symbiotic relationship between hardware engineers, who design boards based on firmware engineer specifications, and the significance of HAL libraries in simplifying the development process.
As the internship progressed, the focus transitioned to programming languages, with C taking center stage. The basics of the language were unraveled through practical exercises that involved creating intricate shapes using asterisks (*), thereby establishing a foundation for subsequent programming challenges. Then, I explored C++, a powerful programming language, and I2C communication protocol. I was tasked with writing a program that interfaced with an ECG sensor, reading and printing its values on the Arduino UNO's serial monitor. Timing charts created using digital oscilloscopes provided invaluable insights into the temporal dynamics of the program execution.

All these mini-projects culminated in working with the engineering team to try to reduce the cost of the device. Specifically, we directed our attention towards refining the ECG circuits within these devices by integrating more efficient microcontrollers and integrated circuit (IC) chips. We evaluated various microcontroller options, considering factors such as processing power, energy efficiency, and cost-effectiveness. Simultaneously, we delved into the realm of IC chips, seeking components that could streamline the ECG circuit's functionality while minimizing costs without compromising on accuracy or performance. We practically implemented these refined circuit configurations, meticulously designing and testing the integration of new microcontrollers and IC chips.
By reducing costs while maintaining or even improving the quality of diagnostics, we were contributing to the overarching goal of HealthCubed: to democratize access to healthcare services. The combination of innovative microcontrollers and IC chips not only held the promise of cost savings but also hinted at enhanced reliability and functionality, reinforcing the value proposition of HealthCubed devices in resource-constrained settings.

Carpl.AI : Accelerating Radiology AI Adoption

The field of medical imaging AI is rapidly advancing, with significant developments in automated image analysis, computer-aided diagnosis, image segmentation and quantification, image reconstruction and enhancement, workflow optimization, multimodal integration, and data augmentation. These advancements have the potential to improve diagnostic accuracy, efficiency, and patient outcomes. AI algorithms can assist radiologists in detecting and diagnosing conditions, provide second opinions, optimize workflow, and enhance image quality.

The internship I undertook at Carpl AI for 3 weeks last winter completely transformed the way I think about the world. Carpl.AI is the world’s first end-to-end platform for the development, testing and deployment of medical imaging AI. Carpl connects AI applications to healthcare providers helping improve access, affordability, and quality of medical care. It onboards AI that small companies make and then has hospitals as clients that can choose from the variety of AI they have onboarded. To use a simple analogy, imagine that Carpl AI is Amazon. The products that amazon has are the different AI models that Carpl has, and the customers - in Carpl’s case, the hospitals - can choose between the different products or AI models. Medical imaging with AI is a relatively new field and new discoveries are being made everyday. At present, it serves as a valuable aid to radiologists, but as its accuracy continues to improve, it holds the potential to assume certain aspects of their role, enhancing overall efficiency in the field.

Throughout my internship, I had the opportunity to work both remotely from Singapore and in person at the Delhi office. Engaging in conversations with professionals from diverse fields within the startup, including technical engineers, digital marketing specialists, and deployment experts, broadened my understanding of the company's operations. Collaborating with Swetha on digital marketing, I conducted comprehensive analyses of Carpl AI's website, Twitter, and LinkedIn presence, juxtaposing them against those of their competitors and collaborators. Notably, I discovered that the website had an impressively high visit duration but also an alarmingly high bounce rate. Leveraging these findings, I provided recommendations to enhance the company's online presence and delivered presentations on website’s best practices and LinkedIn strategies that were subsequently incorporated into their initiatives.

Secondly, I worked with Kritika on the deployment side of the startup. The main areas they work in are in the US, UK, Brazil and Australia. I researched the reimbursement rates for radiologists in these countries and the differences in the public and private healthcare system. I learned a lot from this project about how the healthcare system in different countries works and what they prioritize. Overall, I thoroughly enjoyed my experience at Carpl AI and I learned a lot more than I could have imagined. More than just about what the company does, I learned how to connect with people and what a work environment looks like.

The future of medical imaging AI holds great promise. Advancements will focus on improved accuracy, personalized medicine, and real-time decision support for healthcare professionals. With enhanced algorithms and access to larger datasets, AI systems will offer higher sensitivity and specificity in detecting and diagnosing conditions. Tailoring AI algorithms to individual patients will optimize diagnosis and treatment planning. Additionally, real-time decision support will provide instant insights and flag abnormalities, enhancing efficiency and accuracy in medical imaging interpretation. These developments have the potential to revolutionize healthcare, leading to better patient outcomes and more efficient delivery of care.

One week online + one week in person